gendesign/tradein-mvp/backend/tests/test_same_building_match.py

"""Tests for #6 same-building analog matching via canonical house_id.

Two pieces:
  1. matching.houses.match_house_readonly — read-only target→house resolver
     (cadastr → fingerprint → geo, никогда не INSERT'ит).
  2. estimator._fetch_analogs canonical Tier S — when target_house_id задан,
     матчит listings по house_id_fk; иначе fallback на address-prefix Tier S.
"""
import os

# Settings requires DATABASE_URL at init time. Set dummy DSN before any app import.
os.environ.setdefault("DATABASE_URL", "postgresql://test:test@localhost/test_db")

from datetime import UTC, datetime
from typing import Any
from unittest.mock import MagicMock

from app.services.estimator import _fetch_analogs
from app.services.matching.houses import match_house_readonly


# ── Helpers ───────────────────────────────────────────────────────────────────
def _listing(**over: Any) -> dict[str, Any]:
    base = {
        "source": "cian",
        "source_url": "https://cian.ru/offer/1",
        "address": "Екатеринбург, ул Малышева, д 125",
        "lat": 56.8408,
        "lon": 60.6540,
        "rooms": 2,
        "area_m2": 50.0,
        "floor": 3,
        "total_floors": 9,
        "price_rub": 6_000_000.0,
        "price_per_m2": 120_000.0,
        "listing_date": datetime(2026, 5, 1),
        "days_on_market": 10,
        "photo_urls": [],
        "scraped_at": datetime(2026, 5, 20, tzinfo=UTC),
        "distance_m": 0.0,
        "relevance_score": 0.0,
    }
    base.update(over)
    return base


def _db_returning(rows: list[dict[str, Any]]) -> MagicMock:
    """Session mock where every db.execute().mappings().all() returns rows."""
    db = MagicMock()
    db.execute.return_value.mappings.return_value.all.return_value = rows
    return db


def _executed_sqls(db: MagicMock) -> list[str]:
    return [str(call.args[0]) for call in db.execute.call_args_list if call.args]


# ── match_house_readonly ────────────────────────────────────────────────────
def test_readonly_cadastr_exact_hit():
    db = MagicMock()
    db.execute.return_value.mappings.return_value.first.side_effect = [{"id": 42}]
    result = match_house_readonly(
        db, address="ул Малышева 125", cadastral_number="66:41:0704045:350"
    )
    assert result == (42, 1.0, "cadastr_exact")


def test_readonly_fingerprint_hit():
    db = MagicMock()
    # cadastral_number=None → cadastr tier skipped; first .first() is fingerprint
    db.execute.return_value.mappings.return_value.first.side_effect = [{"house_id": 7}]
    result = match_house_readonly(db, address="Екатеринбург, ул Малышева, д 125")
    assert result == (7, 0.9, "fingerprint")


def test_readonly_geo_proximity_hit():
    db = MagicMock()
    # fingerprint miss → geo hit
    db.execute.return_value.mappings.return_value.first.side_effect = [
        None, {"id": 9, "dist": 12.3}
    ]
    result = match_house_readonly(db, address="неизвестный адрес", lat=56.84, lon=60.65)
    assert result == (9, 0.7, "geo_proximity")


def test_readonly_no_match_returns_none():
    db = MagicMock()
    db.execute.return_value.mappings.return_value.first.side_effect = [None, None]
    result = match_house_readonly(db, address="нигде", lat=56.84, lon=60.65)
    assert result is None


def test_readonly_never_inserts_or_writes():
    """Критично: read-only matcher не должен ничего создавать/писать."""
    db = MagicMock()
    db.execute.return_value.mappings.return_value.first.side_effect = [None, None]
    match_house_readonly(db, address="нигде", lat=56.84, lon=60.65)
    for sql in _executed_sqls(db):
        upper = sql.upper()
        assert "INSERT" not in upper
        assert "UPDATE" not in upper
        assert "ADVISORY" not in upper  # no pg_advisory_xact_lock


def test_readonly_no_coords_skips_geo_tier():
    """Без lat/lon geo-tier не запускается; при fingerprint-miss → None."""
    db = MagicMock()
    db.execute.return_value.mappings.return_value.first.side_effect = [None]
    result = match_house_readonly(db, address="только адрес без координат")
    assert result is None


# ── _fetch_analogs canonical Tier S ──────────────────────────────────────────
def test_canonical_tier_s_fires_with_house_id():
    rows = [_listing(source=s) for s in ("cian", "avito", "yandex", "n1")]
    db = _db_returning(rows)
    listings, _fallback, tier = _fetch_analogs(
        db, lat=56.84, lon=60.65, rooms=2, area=50.0, radius_m=1000,
        full_address="Екатеринбург, ул Малышева, д 125", target_house_id=555,
    )
    assert tier == "S"
    assert len(listings) >= 3
    # Канонический Tier S — первый запрос должен матчить по house_id_fk
    assert "house_id_fk" in _executed_sqls(db)[0]


def test_falls_back_to_address_tier_s_without_house_id():
    rows = [_listing(source=s) for s in ("cian", "avito", "yandex", "n1")]
    db = _db_returning(rows)
    _listings, _fallback, tier = _fetch_analogs(
        db, lat=56.84, lon=60.65, rooms=2, area=50.0, radius_m=1000,
        full_address="Екатеринбург, ул Малышева, д 125", target_house_id=None,
    )
    assert tier == "S"
    # Без target_house_id канонический tier пропущен — первый запрос по address ILIKE
    first_sql = _executed_sqls(db)[0]
    assert "house_id_fk" not in first_sql
    assert "ILIKE" in first_sql


def test_canonical_tier_s_too_few_falls_through():
    """<3 в том же доме → не возвращаем Tier S(canonical), идём дальше."""
    rows = [_listing()]  # только 1 в доме
    db = _db_returning(rows)
    # full_address без распознаваемого short_addr И year/floors None →
    # address-Tier S и Tier H пропустятся, упадём в Tier W (тоже вернёт rows mock).
    _listings, _fallback, tier = _fetch_analogs(
        db, lat=56.84, lon=60.65, rooms=2, area=50.0, radius_m=1000,
        full_address=None, target_house_id=555,
    )
    # canonical Tier S не сработал (1 < 3); tier — не S(canonical) early-return.
    # Mock возвращает те же rows для Tier W, так что итог не пустой, tier='W'.
    assert tier == "W"